test failure: unknown command diff

[Martin Pool]

When I run ' ./bzr selftest -1 -s bt.test_selftest' in trunk, I get

Text attachment: traceback
------------
Traceback (most recent call last):
  File "/home/mbp/lib/python/testtools/runtest.py", line 143, in _run_user
    return fn(*args)
  File "/home/mbp/lib/python/testtools/testcase.py", line 539, in _run_test_method
    return self._get_test_method()()
  File "/home/mbp/bzr/work/bzrlib/tests/test_selftest.py", line 3043, in test_test_suite
    suite = tests.test_suite()
  File "/home/mbp/bzr/work/bzrlib/tests/__init__.py", line 4005, in test_suite
    plugin_suite = plugin.test_suite()
  File "/home/mbp/bzr/work/bzrlib/plugin.py", line 562, in test_suite
    return self.module.test_suite()
  File "/home/mbp/.bazaar/plugins/bzrtools/__init__.py", line 89, in test_suite
    import tests.shelf_tests
  File "/home/mbp/.bazaar/plugins/bzrtools/tests/shelf_tests.py", line 11, in <module>
    from bzrlib.plugins.bzrtools.command_classes import cmd_shelf1
  File "/home/mbp/.bazaar/plugins/bzrtools/command_classes.py", line 531, in <module>
    class cmd_cdiff(BzrToolsCommand):
  File "/home/mbp/.bazaar/plugins/bzrtools/command_classes.py", line 534, in cmd_cdiff
    takes_options = list(get_cmd_object('diff').takes_options) + [
  File "/home/mbp/bzr/work/bzrlib/commands.py", line 240, in get_cmd_object
    raise errors.BzrCommandError('unknown command "%s"' % cmd_name)
BzrCommandError: unknown command "diff"
------------

[Aaron Bentley (abentley)]

This seems more like a bug in bzr. If you want to make things lazy, go ahead. But when I ask for something by name, it should return it, don't you think?

[Martin Pool (mbp)]

It may well be a bzr bug, and it probably is caused by changing to
lazy registration of commands. I haven't dug into it yet.

It does seem like it would be inefficient to make every test that
wants builtin commands exist have to say so, if that's what has
happened.

[John A Meinel (jameinel)]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On 4/15/2011 11:46 PM, Martin Pool wrote:
> It may well be a bzr bug, and it probably is caused by changing to
> lazy registration of commands. I haven't dug into it yet.
>
> It does seem like it would be inefficient to make every test that
> wants builtin commands exist have to say so, if that's what has
> happened.
>

Note that bzrtools adds a test into test_selftest, that fails when diff
is not available as an executable.

John
=:->
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (Cygwin)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/

iEYEARECAAYFAk2p9hYACgkQJdeBCYSNAAOCGACcDHeZIyxRNJ2OEqwf2l971NDS
w2MAn0gQSXitXWU6sbm6pv74cn4k/XZy
=4eaw
-----END PGP SIGNATURE-----